Rachmat, Jauri Bima Arli Nur and drh. Widi Nugroho, and Bayu Rahayudi, (2022) Evaluasi Sensitivitas Dan Spesifisitas Deteksi Fasciola Menggunakan Kecerdasan Buatan Object Detection Yolov4 Terhadap Sampel Laboratorium Telur Fasciola Sp. Segar Dan Awetan Formalin. Sarjana thesis, Universitas Brawijaya.
Abstract
Fascioliasis pada hewan ternak, merupakan penyakit yang umum ditemui dan menjadi permasalahan yang serius. Fascioliasis dapat menyebabkan penurunan produktifitas susu, penerunan berat badan, menurunya kualitas karkas, hingga kematian hewan ternak. Prevalensi fasciolosis di indonesia cukup tinggi berkisar 40-90%. Terdapat berbagai macam teknik diagnosa yang dipakai pada saat ini, antara lain, pemeriksaan feses, pemeriksaan serologi, biopsi hati, dan ELISA. Beberapa teknik tersebut masih dilakukan secara manual sehingga rawan terjadinya kesalahan membaca hasil, dengan perkembangan teknologi yang semakin maju teknik pemeriksaan khususnya dalam pemeriksaan feses dapat dilakukan, secara otomatis salah satunya algoritma YOLOv4. Tujuan penelitian ini adalah untuk mengembangkan diagnosa fasciolosis pada sapi menggunakan kecerdasan buatan, untuk mempermudah dan mempercepat proses diagnosa. Pada penelitian ini digunakan sampel telur Fasciola sp. segar dari empedu sapi yang mengalami fasciolosis dan awetan formalin dari gerusan cacing Faciola sp. dewasa. Sampel kemudian diamati dengan mikroskop dan diambil gambar menggunakan kamera ponsel. Data citra telur pada penelitian ini digunakan untuk membangun diagnosa object detection. Pada penilitan ini menggunakan 55 gambar dengan 25 gmabar Fasciola sp. segar, 25 Fasciola sp. awetan formalin, dan 5 gambar gabungan keduanya. Proses object detection dilakukan menggunakan algoritma deep learning dengan alat bantu YOLOv4, luaran dari object detection yang dilakukan adalah untuk mengetahui sensitivitas dan spesifisitas. Proses training, validasi, dan testing menggunakan cloud computing Google Colaboratory. Diperlukan Proses validasi menggunakan cross validation. Hasil penelitian menunjukkan nilai sensitivitas dan spesifisitas masing-masing adalah 100% dan 30% sehingga algoritma ini mampu mengenali telur Fasciola sp segar namun belum mampu secara maksimal membedakan antara Fasciola sp. segar dan awetan formalin. Algortitma ini masih belum bisa digunakan langsung di lapangan dan perlu dikembangangkan lebih lanjut. Kata kunci: Fascioliasis, Kecerdasan Buatan, Uji Sensitivitas, Uji Spesifisitas, YOLOv4.
English Abstract
Fascioliasis in livestock is a common disease and is a serious problem. Fascioliasis can cause decreased milk productivity, decreased body weight, decreased carcass quality, and even death of livestock. The prevalence of fasciolosis in Indonesia is quite high, ranging from 40-90%. There are various diagnostic techniques used today, including stool examination, serological examination, liver biopsy, and ELISA. Some of these techniques are still carried out manually so that they are prone to errors in reading the results. With the development of increasingly advanced technology, examination techniques, especially in the examination of feces, can be carried out automatically, one of which is the YOLOv4 algorithm. The aim of this research is to develop a diagnosis of fasciolosis in cattle using artificial intelligence, to simplify and speed up the diagnosis process. In this study, Fasciola sp. egg samples were used. fresh from the bile of a cow experiencing fasciolosis and preserved in formalin from the scouring of Faciola sp. mature. The samples were then observed under a microscope and pictures were taken using a cellphone camera. Egg image data in this study is used to develop object detection diagnostics. In this research using 55 images with 25 images of Fasciola sp. fresh, 25 Fasciola sp. preserved formalin, and 5 combined images of both. The object detection process is carried out using a deep learning algorithm with the YOLOv4 tool, the output of the object detection is to determine the sensitivity and specificity. The training, validation and testing process uses Google Colaboratory cloud computing. Required validation process using cross validation. The results showed that the sensitivity and specificity values were 100% and 30%, respectively, so that this algorithm was able to recognize fresh Fasciola sp eggs but was not able to maximally distinguish between Fasciola sp. fresh and preserved in formalin. This algorithm still cannot be used directly in the field and needs to be developed further. Keywords: Fascioliasis, Artificial Intelligence, Sensitivity Test, Specificity Test, YOLOv4
Other obstract
-
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | 052213........ |
| Subjects: | 600 Technology (Applied sciences) > 636 Animal husbandry > 636.08 Specific topics in animal husbandry > 636.089 Veterinary medicine |
| Divisions: | Fakultas Kedokteran Hewan > Kedokteran Hewan |
| Depositing User: | Nur Cholis |
| Date Deposited: | 28 Aug 2023 01:44 |
| Last Modified: | 28 Aug 2023 01:44 |
| URI: | http://repository.ub.ac.id/id/eprint/202645 |
![[thumbnail of DALAM MASA EMBARGO]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text (DALAM MASA EMBARGO)
Jauri Bima Arli Nur Rachmat.pdf Restricted to Registered users only until 31 December 2024. Download (2MB) |
Actions (login required)
 |
View Item |